Re: Proposing _NET_WM_STATE_FULLSCREEN_EXCLUSIVE

On Fri, Nov 2, 2012 at 3:08 PM, James Cloos <[email protected]> wrote:
>>>>>> "MG" =3D=3D Martin Gr=E4=DFlin <[email protected]> writes:
>
> MG> Now if a game changes the resolution this gets completely destroyed.
>
> The bug is not that full-screen apps might want to change the resolution,
> it is that randr doesn't keep the virtual size constant when that occurs.
>

...

> The problem is that using randr to change the visual size changes the
> x11 screen's dimensions, too, with fallout for WMs.  The bug needs to
> be fixed either in xrandr(1) or in the X server's implmentation of the
> randr spec.
>
> -JimC

I disagree; we no longer have a world of CRTs (where scaling is an
easy analog result of device function) but a world of exactly
addressed pixels.  The exact shape and patterning of these pixels
differs, but there is a correct 'native' resolution with which to
match a display's capability. The reason that 'performance' is taken
on the GPU is a conscious choice that quality instead of raw speed is
desired;less accurate algorithms might be used which would provide
faster but less accurate results.  There's also the flexibility to do
other things at the same time.

I do agree on being more precise about what you are requesting of the
window management stack.  My firm belief is that this must be a
request, unless the application has no window manager to ask it should
never adjust those settings.